public-commons-media-search

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). At runtime, the required actions (e.g., search_titles, list_page_media, search_commons_media, get_file) fetch and return Wikipedia/Wikimedia Commons page excerpts/snippets and file descriptions (outsider-authored public web content) into the agent’s tool response, which is then placed into the LLM context.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill explicitly instructs runtime calls to the AgentPMT MCP endpoint (https://api.agentpmt.com/mcp/) — e.g., using the agentpmt-tool-search-and-execution actions get_schema/get_instructions — to fetch live schema/instructions which would directly supply runtime prompts/instructions to the agent.